Realistic virtual reproductions. Image-based modelling of geometry and appearance.

摘要

Existing technologies for contact-less 3D scanning and Image Based Modelling (IBM) methods are being extensively used nowadays to digitize cultural heritage elements. With a convenient degree of automation these methods can properly capture and reproduce shape and basic colour textures. However, there is usually a quite evident lack of fidelity in the resulting appearance of the virtual reproductions when compared with the original items. Even when properly photo-textured, the reproduced surfaces often resemble either plaster or plastic, regardless of the properties of the original materials. What is neither captured nor modelled is the natural dynamic response of the actual materials with respect to changes in observation angle and/or the lighting arrangement. The methodology introduced in this paper tries to improve the three-dimensional digitalization and visualization of cultural heritage elements, by extending the present capabilities of IBM with additional capture and modelling of surface appearance. We show that it is possible to automatically reproduce realistic-looking virtual objects and scenes, even with photographs taken with an uncalibrated single moving camera and while under uncontrolled and intentionally variable lighting conditions. This is achieved not only by reconstructing the shape and projecting colour texture maps from photographs, but also modelling and mapping the apparent optical response of the surfaces to light changes, while also determining the variable distribution of environmental illumination of the original scene. This novel approach integrates Physically Based Render (PBR) concepts in a processing loop that combines capture and visualization. Using the information contained in different photographs, where the appearance of the object surface changes with environmental light variations, we show that it is possible to enhance the information contained in the usual colour texture maps with additional layers. This enables the reproduction of finer details of surface normals and relief, as well as effective approximations of the Bi-directional Reflectance Distribution Function (BRDF).